Control With Delay: Why Brains Predict

A live demonstration of feedback, instability, internal models, anticipatory control, and learning from prediction error.
Supertanker in a narrow channel Error: small Swings: none
WATCH
Stability trace
actual delayed sensory state model estimate target lane
Phase plot error × velocity
Prediction error: the teaching signal delayed senses minus what the model predicted for that same moment
What the model believes
ship's inertia model's inertia
Overshoot on each lane change
Yellow actual now Gray sensed past Cyan estimated now Green target lane Magenta prediction error ↕Blue gust or current ↕Orange corrective action
Watch: current → correction → motion; delay separates gray from yellow.
Keyboard: Space run/pause · 1–6 presets · ↑/↓ gusts · R reset. The simulation is intentionally simplified: it is a demonstration of delayed feedback and model-based control, not a navigational model of a particular ship or a literal model of the cerebellum.